MWP – Artificial Intelligence Channel cover art

Artificial Intelligence

Episodes whose main subject is AI itself: building with, training, evaluating, or debating AI systems. AI used only as a tool for another topic goes to that topic.

1594 episodes Page 11 of 80 RSS Feed All feeds

#6272: Teaching Audio Models to Hear Beyond Words

Speech recognition rewards ignoring background noise, but audio-language models train specifically on environmental sounds and temporal scenes.

speech-recognitionlocal-aiai-evaluationai-inferencetraining-data

#6271: Why Video Search Often Works Better as Text

A one-minute video compressed into a single vector undergoes roughly five orders of magnitude of raw data reduction.

context-and-memorycomputer-visionspeech-recognitionragai-evaluation

#6267: Why Text-to-Speech Fails at Explaining Code

A study comparing Google TTS, ChatGPT, Gemini, ElevenLabs, and Copilot found every system failed to communicate indentation, scope, and code comments.

speech-recognitionprogramming-languagestext-to-speech

#6266: Balancing Speech Training Data With Square Roots

A fine-tuning recording app weighs tasks by the square root of unread sentences so massive datasets don't crowd out smaller prompt sets.

speech-recognitiontraining-datafine-tuning

#6265: Why Upgrading Embedding Models Demands Reindexing

Cohere's Embed Five Fast queries against a Pro index reach 98.4 percent of performance without regenerating stored document vectors.

context-and-memorylocal-aiai-inferenceragtraining-data

#6264: Choosing Omni Embedding Models for Mixed Media

Jina's omni model trains only 0.35 percent of its weights while aligning with existing text indices.

speech-recognitioncontext-and-memorylocal-airagtraining-data

#6263: Why Numeric Data Lacks a Default Embedding Model

Popular time-series models like IBM's Granite TSPulse forecast trends rather than provide canonical vectors for similarity search.

context-and-memoryragtraining-dataai-inference

#6257: Where Your Prompts Actually Go

Selecting US-only inference on Anthropic's direct API incurs a 1.1x multiplier across token categories, while global routing is the standard rate.

cloud-computingai-inference

#6255: Routing Claude Code Tasks to Local Ollama Models

A feature request to support local subagent endpoints in Claude Code was closed for inactivity after less than two months.

ai-coding-assistantsmodel-context-protocolcyberattacksmulti-agent-systemslocal-ai

#6254: Where a Synthetic Voice Gets Its Accent

A neural speech interface maps motor cortex signals into phonemes drawn from an eclectic training set of human voice recordings.

speech-recognitionai-alignmenttext-to-speechai-biasmathematics

#6253: Why Context Windows Stalled at a Million Tokens

A dense 70B model requires about 328 gigabytes of KV cache to serve a single one-million-token prompt, exceeding the memory of an Nvidia B300 chip.

context-and-memoryai-inferencelocal-aitransformer-architecturegpus

#6245: Why Inference Engines Change Benchmark Scores

A study found a 16.6 percentage point spread on GSM8K across different inference engines running identical weights on the same GPU.

ai-inferencelocal-aiai-evaluationopen-sourcetransformer-architecture

#6244: When Your Fine-Tune Isn't Done Learning Yet

Falling loss, 90% word error rate, and a model that still says "Derry gym lag." What undertraining actually looks like — and when more epochs won't...

speech-recognitionai-evaluationfine-tuningtraining-datalocal-ai

#6243: Inside the Mixture of Experts Router

What actually decides which expert handles your token? A look inside the router, from sigmoid affinities to load balancing.

ai-inferencetransformer-architectureopen-sourcegpus

#6242: Training a Model on Your Own Taste

What does it actually take to train a small model on your editorial taste using pairwise preferences and DPO?

prompt-engineeringai-alignmentfine-tuninglocal-aitraining-data

#6241: Why Your Synthetic Data Isn't as Diverse as You Think

Fifty-nine percent of one GPT-4 dataset started with "Once upon a time." Correct examples can still make a terrible training set.

ai-evaluationtraining-dataprompt-engineeringfine-tuning

#6240: Gold vs Silver Training Data: What Makes Labels Trustworthy

Human-checked "gold" labels and machine-made "silver" ones aren't as different as the names suggest — and the gap matters.

ai-evaluationtraining-datahallucinationsfine-tuning

#6239: Testing Synthetic Data Generators Before You Scale

A benchmark trains student models on each generator's output to find out which model actually writes useful training data.

ai-evaluationtraining-datahallucinationsfine-tuning

#6238: Which AI Models Actually Generate Good Training Data?

Open weights vs closed APIs for synthetic data, and why IFEval scores don't tell you which model will hold up.

ai-evaluationtraining-datafine-tuninglocal-aiai-inference

#6237: Small Models, Big Questions: When Cheap Fables Get Expensive

Three million fables for pennies — but can a small model tell a good example from a plausible one?

ai-evaluationlocal-aiai-inferencetraining-datafine-tuning